A Research Query, from Victor Niederhoffer
What is the best formula for description, and separately for prediction, for separate markets (say S&P, bonds, gold, …) involving the first, second and third difference, taking account the number of past observations considered such as last 4, 5, 7 or 10.